Cancer survival is now shown in ten 5-year periods from 1965-1969 to 2010-2014.
In version 7.0 to 7.2 (December 2014 to July 2016) the survival estimates for Sweden were underestimated, in particular for prostate, brain, kidney, leukaemia and small intestines cancers for the years 1974-1993. The reason for this was inclusion of cases only known from autopsies. Such cases are usually excluded from cancer survival estimation. For the group All sites but non-melanoma skin cancer 1- and 5-year relative survival were up to 5% point too low for cases diagnosed in 1974-1993. Since then, the proportion of cancers found by autopsy alone has decreased in Sweden and from 1999 the underestimation has been 1-2% point. The error has now been rectified, leading to an overall improvement in Swedish cancer survival.
In version 6.0 and 6.1 (November 2013 to December 2014) the relative survival figures shown for the years 2009-2011 for Finland, Norway and Sweden were too high, 1-year survival on average 2.5% points and 5-year survival 5% points too high. Figures are now corrected.
The Swedish cancer registry does not search for and perform follow-back in other registers for cancer cases identified from death certificates, as the other Nordic cancer registries do. This may lower Swedish incidence rates, especially for older ages and for cancers types with a high lethality, and thus result in too optimistic survival, since incident cases with short survival may not be included. Lung and pancreatic cancer are examples of cancer types where Swedish incidence rates are not directly comparable to rates from other Nordic countries and the use of mortality rates for comparisons is recommended.
Excluding cases first identified via a cancer cause of death registration from the Danish cancer data 1999-2003 increased the 1-year relative survival by 1.5% points or less.
More detailed warnings can be found under Notes in “The NORDCAN database”.